Chess is different when it comes to forms of artificial intelligence. We use chess engines all the time (except when playing serious games) and using them to answer questions is so normal that it is more usual to make the excuse "I didn't use an engine" to highlight that any errors in presented analysis are human not silicon ones.
ChatGPT, however, is entirely different. The answers to chess questions are banal and useless. They are usually, at best, of the form "God, motherhood and apple pie are good". At their worst they are just silly and wrong. Here is one example in response to this question. Actually, that answer is just plain wrong and illustrates just how incompetent ChatGPT is when it comes to chess.
The overall site policy is given in this Meta post. It is that individual sites must decide their own specific policy as to whether such answers are acceptable but that in any case all such answers must include attribution to ChatGPT because they are not the author's own thoughts.
What do people think about this? Should we allow ChatGPT with attribution? Should we ban it? Should we allow in only certain special cases?
FWIW My view is to ban ChatGPT altogether while strongly encouraging the use of real chess-based artificial intelligence like Stockfish. LeelaZero, et al. Even GNUChess and Crafty.